Period averages of 10 years were used in the construction of the panel. Non-overlapping 10-year periods starting from 1960 to 1999 were utilized, hence a maximum of 4 observations per country (i.e., 1960-1969; 1970-79…1999). A sample of 36 SSA countries was selected for the study. Two dummies, high and medium primary commodity dependence, were introduced. The dummies took on the value of 1 for countries that exhibited high and medium commodity dependence respectively, and 0 for otherwise.
Countries that obtained more than 70% of their export earnings from three main primary commodity exports were designated as high commodity-dependent countries (Hdep) and those that had less than 70% as medium-dependent (Mdep) (See Appendix Tables 1 and 2).
Population figures in the various countries were used as proxies for labour participation in the economies. Gross capital formation was the other independent variable. The value of exports was assumed to be equal to earnings from primary commodity exports for lack of data on primary commodity exports. In any case, as indicated in the earlier section on stylized facts about SSAs exports, the share of manufactures in merchandise trade was quite insignificant (UNCTAD, 2003).
The literature indicates an absence of a well-defined measure of export revenue instability (Gyimah-Brempong, 1991). In the present study, however, the Holdrick-Prescott Filter was used in the construction of the export instability index (Holdrick & Prescott, 1997).
First, the export series was decomposed into its trend and cyclical components respectively. Normalized deviations from the trend were then used as a measure of instability. All the data used in the construction of the dependent and independent
variables were obtained from the World Development Indicators, WDI (World Bank, 2004).
Summary statistics
The descriptive statistics for our panel data set presented in the Table below indicate that SSA experienced very low real per capita GDP growth over the past 40 years. Per capita GDP in SSA on the whole grew at an average rate of 0.8% per annum and varied from – 6.4% to 17.3% per annum. A cursory look at the other explanatory variables also indicates some weaknesses that appear not to have encouraged strong growth. Export performance across the region reveals a grim picture. On average SSA exports grew by barely 11% per annum over the period 1960-1999. Growth in exports varied from -20% per annum to 15%.
Export instability, on the other hand, varied from 17% below trend to less than 1% above trend (see normalised instability indices in Table 2-2). It can therefore be inferred that primary commodity export instability has mostly been on the negative side (below trend).
The paradox, however, is the fact that in the midst of such unfavourable instability certain countries in SSA such as Botswana and Ghana have recorded steady growth (modest for Ghana but strong for Botswana) over the past two decades. Yet countries such as Zambia and Nigeria, among others, have not seen much stability and resilience in per capita growth. Those countries in the less resilient group have seen their fortunes fluctuate alongside export earnings.
Table 2-2. Summary Statistics of common Panel (N=89), 1960-1999
Statistic
Standard Deviation 3.075 0.190 1.566 11.188 0.007 Coefficient of variation 3.688 1.781 0.328 1.923 0.272 Maximum 17.247 0.150 0.007 56.695 0.045
Minimum -6.436 -0.198 -0.167 -23.605 0.001 Median 0.547 0.067 -0.168 4.558 0.026
2.6 Results
The econometric estimation was begun by experimenting with pooled-OLS, OLS-differences and Least Square Dummy Variable (LSDV) estimators. Within, Between and Maximum Likelihood estimators were also used in the estimation of the empirical model.
An inspection of the estimations of the various coefficients, the signs and other robustness criteria indicated the superiority of the OLS-differences estimator. Results of the model estimated with the OLS-difference estimators are presented in Table 2-3. The first column of the Table presents estimates of the base model. Per capita GDP growth rate, Yi,t, was the dependent variable, whilst changes in population Li,t, exports Xi,t and changes in gross capital formation Ki,t constituted the regressors.
In order to draw a meaningful conclusion from the empirical work a base model is used as a benchmark along which the substantive model for the investigation is assessed. In addition to the base model two other models were estimated. These were essentially the base model augmented with two different regressors of importance. The variables that were used in the subsequent and systematic augmentations were export proceeds instability index (INSTi,t)and dummies that captured the extent of dependence on primary commodities. The dummies were Hdepi,t and Mdepi,t for high and medium commodity dependence respectively. These take on unity for high and medium dependence and 0 for otherwise.
The coefficient estimates, standard errors of the estimates and a select range of statistics are presented in the Table 2-3. The models progressively explain reasonably large
proportions of the variations in economic growth performance across countries in SSA.
The highly significant Wald Chi2 statistic (an analogue of the F-test), which tests the joint significance of the explanatory variables, appears to be significant at the 99% level of significance for all 3 models. The value of the statistic also increases across the models steadily from 21.53 for the base model, and to 21.69 and 28.16 for models 2 and 3 respectively. The apparently high R2 for the models – given that panel data estimations usually have characteristically low R2 values – further underscores the usefulness of the selected models. Thus the selected model fits the data quite well.
The base or control model explains a reasonably high proportion of the variation in economic growth across SSA. The coefficients of all the explanatory variables have signs that conform to a priori expectations. Changes in gross capital formation in particular have a positive sign and a coefficient (with t-probability of 0.002) indicating economic growth is strongly related to gross capital formation in SSA. Export, the other significant regressor with t-probability of 0.028 indicating significance at 95%, was found to have the right sign.
This result is consistent with earlier studies that find a positive relationship between exports and economic growth (Feder, 1983; Kreuger et al., 1980 and Gyimah-Brempong, 1991). Intuitively this appears right, since most SSA countries depend on a narrow range of commodity exports to finance the imported inputs for production.
Even though the labour coefficient had the right expected sign, this was not statistically significant. The fact that the coefficients of growth in exports and gross capital formation were significantly different from zero [at least at α =0.05] indicates that economic growth in SSA is positively related to growth rates of gross capital formation and exports. These results are consistent with neoclassical growth theory and previous results of studies on growth in LDCs. The key objective of the present paper, as stated earlier, was to ascertain
empirically the relationship between export earning instability and economic growth.
Consequently we now turn to discussions of models 2 and 3. These are basically the base model which has been augmented by an instability and dependency index and a dummy respectively. When the instability measure is added to the base model, the explanatory power of the model as measured by R2 and the Wald Chi2 statistics, improves significantly (see Table 2-3).
More importantly, the coefficient of export instability is negative, but not significantly different from zero at α = 0.10. Again, when the instability indices were added to the base model, the signs and coefficients did not vary significantly; however, the adjusted R2 and the Wald Chi2 statistics improve modestly from 0.43 and 21.53 to 0.45 and 21.69 respectively.
Table 2-3. Modelling per capita GDP growth, Yi,t using OLS-differences estimator
Base Model Model 1 Model 2 Variable
Coefficients
Exports, Xi,t 19.737 22.100 21.414
(8.494) (9.172) (8.808)
Capital, Ki,t 0.139 0.146 0.146
(0.040) (0.041) (0.039)
Labour, Li,t 33.919 39.89 23.455
(86.20) (87.22) (90.220)
Instability, INSTi,t - -5.496 -7.872
(7.616) (7.412)
High dependence, Hdep i,t - - -3.286
(1.584)
Medium dependence, Mdepi,t - - -2.410
(1.603)
Constant, bo -0.143 -0.154 -0.187
(0.039) (0.393) (0.377)
Adjusted R2 0.434 0.445 0.529
Wald Chi2 21.53** 21.69** 28.16**
Observations 89 89 89 Notes: Econometric analysis was carried with PCGive 10.
This underscores the fact that introduction of the instability measure do not necessarily introduce any meaningful biases to the coefficient estimates nor does it introduce multi-collinearity.
The results of the coefficient estimates of the new model (i.e. model 3) lend empirical support to the significance of primary commodity dependence to growth outcomes in SSA countries. The signs of the two dummies were negative, as expected. However, the coefficient of medium commodity dependence, unlike high commodity dependence, was not significantly different from zero. The significant improvements in the test statistics R2 and the Wald Chi2 from 0.45 and 21.69 to 0.53 and 28.16 respectively further enhance the capability of the model to explain variations in real per capita growth rates in SSA countries. Better still, when model 3 is put beside the base model, marked improvements in the evaluation test statistics are seen.
2.7 Conclusions
This paper ascertained whether export-earnings instability has any significant effect on economic growth performance in SSA countries. Also examined was the impact of varying degrees of commodity dependence on economic growth. The study used a panel data framework and data on a sample of 38 SSA countries with estimation period 1960 to 1999. The key finding is that there exist a negative relationship between commodity export instability and economic growth but this was statistically insignificant. However, the relationship between extent of commodity dependence and economic growth was negative and statistically significant.
The results indicate the need for a diversification of the export base of these countries in the short to medium term. In the long term, however, deliberate efforts need to be directed at diversifying exports to include manufactures. Hitherto conventional wisdom has pushed for increased commodity exports but since commodity price changes results in export revenue instability efforts need to made to move away from high commodity dependence. Development policies need to be aimed at manufacturing and service sector export-led growth. Nonetheless, it’s not suggested that all countries on the continent follow such a prescription. This is especially true for those parts of Africa that are landlocked and may not necessarily be successful in export-led growth in manufactures in low-wage industries, because of the high transport costs involved.
List of References
Adams, F.G., Behrman, J. R. & Roldan, R. A., 1979. Measuring the impact of primary commodity fluctuations on economic development: coffee in Brazil. American Economic Review, 69 (2), p.164-168.
Akiyama, T., J. Baffes, Larson, D. F. & Varangis, P., 2003. Commodity market reform in Africa: some recent experience. World Bank Policy Research
Working Paper, 2995. Washington D.C.: World Bank.
Baffes, J., 2001. Policy reform experiences in cotton markets, in: Akiyama, Baffes,
Larson and Varangis, eds., Commodity market reforms: lessons of two decades. Washington, D.C.:
World Bank.
Collier, P., 2002. Primary commodity dependence and Africa’s future. Washington, D.C.: World Bank.
Cuddington, J.T. & Urzua, C.M., 1989. Trends and cycles in the batter trade net terms of trade: a new approach. Economic Journal, 99, p.426-442.
Cuddington, J.T & Liang, H., 1998. Commodity price volatility across exchange rate regimes.
Washington, D.C.: Department of Economics, George Town University.
Coulson, N. & Robins, R., 1985. Aggregate economic activity and the variance of inflation: another look. Economic Letters, 17, p. 71-75.
Feder, G., 1983. On exports and growth. Journal of Development Economics, 12, p. 59-73.
Glezakos, C., 1984. Export instability and economic growth: reply. Economic Development and Cultural Change, 32, p. 229-36.
Gyimah-Brempong, K., 1991. Export instability and economic growth in Sub-Saharan Africa. Economic Development and Cultural Change, 39, p. 815-28.
Hendry, D.F. & Doornik, J. A., 2001. PCGive 10, an interactive econometric modelling system.
London: Timber Lake Consultants Press.
Hodrick, R. & Prescott, E. C., 1997. Postwar U.S. business cycles: an empirical investigation. Journal of Money, Credit, and Banking, 29(1), p.1-16.
IMF, 2005. International Financial Statistics Year Book. [CD-Rom]. Washington, D.C:
International Monetary Fund.
Kenen, P.B. & Voivodas, C., 1972. Export instability and economic growth. Kyklos 25, p.
791-801.
Knudsen, O & Parnes, A., 1975. Trade instability and economic development. London: D.C.
Heath & Co.
Kreuger, A. O., Schiff, M. & Valdes, A., eds., 1991. The political economy of agricultural pricing policy. Baltimore: Johns Hopkins University Press.
Lam, V.M., 1980. Export instability and primary commodity concentration. Economia Internazionale, 33 (1), p. 40-57.
Larson, D., Varangis, P. & Yabuki, N., 1998. Commodity risk management and
development. World Bank Policy Research Working Paper 1963. Washington, D.C.: World Bank,
Love, J., 1992. Export instability and the domestic economy: questions of causality. Journal of Development Studies, 28(4), p.735-42.
MacBean, A. I., 1966. Export instability and economic development. London: Allen & Unwin.
Maizels, A., 1968. Review of export instability and economic development by MacBean. American Economic Review, 58, pp 575-580.
Mullor-Sebastian, A., 1988. Export fluctuations and economic growth. Economic Development and Cultural Change, 36 (3), p. 3:217-36.
Mundlak, Y.,Cavallo, D. & Domench, R., 1993. Agriculture and economic growth in Argentina, 1913-84, Research Report No. 76, Washington, D.C.: International Food Policy Research Institute.
Ozler, S & Harrigan, J., 1988. Export instability and growth. Department of Economics Working Paper, No. 486, Los Angeles: University of California.
Obidegewu, C.F & Nzimasanga, M., 1981. Copper and Zambia: An econometric Analysis. Lexington, Mass: Lexington Books, D.C. Heath & Co.
Priovolos, T., 1981. Coffee and the Ivory Coast: an econometric study. Lexington, Mass:
Lexington Books, D.C. Heath & Co.
Savvides, A., 1984. Export instability and economic growth: some evidence.
Economic Development and Cultural Change, 32, p.607-14
Sinha, D., 1999. Export instability, investment and economic growth in Asian countries: a time series analysis. Economic Growth Centre Discussion, Paper, No. 799. Yale University and Macquarie University, Australia.
Yotopoulus, P. A & Nugent, J. B., 1977. Economic development: empirical investigations. New York: Harper & Row.
UNCTAD. 2003. Economic development in Africa: trade performance and commodity dependence.
Geneva: United Nations Conference on Trade and Development, United Nations.
Varangis, P. & Schreiber, G., 2000. Cocoa market reforms in West Africa: policy challenges. World Bank Technical Paper No. 444. Washington, D.C.: World Bank
Wilson, P., 1994. Export earnings and instability in Singapore, 1957-1988: a time series analysis. Journal of Asian Economic, 5, p.399-412.
World Bank, 2004. World development indicators. [CD-Rom]. Washington, D.C.: World Bank.
APPENDIX
The sample of 36 SSA countries for the study was grouped into 3 sub-samples. The grouping was based on the level of primary commodity concentration. Level of primary commodity concentration was defined using the proportion of national revenue that was obtained from a given country’s 3 key primary commodity exports. Period average earnings from primary commodity exports for 1990-1999 were used (UNCTAD, 2003). Countries that obtained 70% or more of foreign exchange earnings from their 3 major primary commodity exports were classified as high primary commodity concentration countries. Those that obtained between 30-69% were designated medium commodity concentration countries and for less than 30% low concentration (see Table A1). Table A2 describes the degree of commodity dependence generally in SSA
Table A1. Measure of primary commodity diversification High primary
Botswana Mali Mauritius
Burundi Cote d'Ivoire Swaziland
Cameroon Seychelles Lesotho
Central Africa Republic Kenya
Chad Madagascar Congo, Rep. Senegal
Equatorial Guinea Burkina Faso
Ethiopia Zimbabwe Gabon Gambia
Ghana Benin Guinea Sudan Guinea Bissau Cape Verde
Malawi
Table A2. Primary Commodity Dependence No. Country Average 90-99
(%) Leading commodity exports 1 Botswana 95 Diamonds sorted, bovine meat, hides and skin 2 Niger 94 Uranium, live animals, tobacco
3 Gabon 92 Oil, timber, manganese ore
No. Country Average 90-99
(%) Leading commodity exports 4 Congo, Rep. 91 Oil, timber, sugar
5 Congo, Democratic Rep. 89 Diamonds, coffee, timber 6 Nigeria 87 Oil, cocoa, rubber
7 Comoros 87 Vanilla, essential oils, cloves 8 Burundi 87 Coffee, tea and sugar 9 Equatorial Guinea 83 Oil, timber, cocoa 10 Guinea Bissau 82 Nuts, fish, cotton lint 11 Sao Tome and Principe 81 Cocoa, fish, coffee
12 Ethiopia 80 Coffee, hides and skins, sesame seeds 13 Angola 80 Fuels, diamonds sorted, coffee
14 Malawi 76 Tobacco, tea, sugar
15 Mauritania 76 Iron ore, fish, oil
16 Central Africa Republic 70 Diamonds sorted, timber, cotton lint 17 Uganda 68 Coffee, fish, crude metals
18 Zambia 68 Copper, sugar, cotton lint 19 Togo 66 Phosphate, cotton lint, coffee 20 Rwanda 65 Coffee, tea, hides and skin 21 Cameroon 62 Oil, timber, cocoa
22 Chad 62 Cotton lint, cattle, crude metals 23 Guinea 62 Bauxite, alumina, fish
24 Ghana 61 Cocoa, diamonds, gold
25 Mali 58 Cotton lint, live animals, oil of groundnuts 26 Cote d'Ivoire 56 Cocoa, oil, coffee
27 Seychelles 55 Fish, oil, cinnamon oil 28 Somalia 51 Live animals, bananas, fish 29 Namibia 50 Diamonds, fish, live animals 30 Mozambique 49 Fish, nuts, timber
31 Kenya 46 Tea, coffee, oil 32 Madagascar 45 Fish, coffee, cloves 33 Senegal 45 Fish, oil, groundnut oil
34 Burkina Faso 39 Cotton lint, sesame seed, hides and skin 35 Zimbabwe 36 Tobacco, cotton, gold
36 Gambia 35 Groundnuts, fish, groundnut oil 36 Benin 33 Cotton lint, cotton seed, oil of pal, 37 Sudan 32 Sesame seed, crude metals, coarse grains 38 Cape Verde 30 Fish, apples, timber
39 Mauritius 27 Sugar, crude metals, fish 40 Sierra Leone 23 Fish, coffee, cocoa
41 Swaziland 22 Sugar, citrus fruits, other fruits 42 Liberia 18 Rubber, timber, oils
Source: UNCTAD, 2003